CN111508532B - Health management for magnetic storage media - Google Patents
Health management for magnetic storage media Download PDFInfo
- Publication number
- CN111508532B CN111508532B CN202010076783.7A CN202010076783A CN111508532B CN 111508532 B CN111508532 B CN 111508532B CN 202010076783 A CN202010076783 A CN 202010076783A CN 111508532 B CN111508532 B CN 111508532B
- Authority
- CN
- China
- Prior art keywords
- storage medium
- read
- magnetic storage
- region
- metrics
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000036541 health Effects 0.000 title claims abstract description 248
- 238000013528 artificial neural network Methods 0.000 claims abstract description 73
- 238000000034 method Methods 0.000 claims description 44
- 230000015654 memory Effects 0.000 claims description 38
- 230000006870 function Effects 0.000 claims description 14
- 230000007774 longterm Effects 0.000 claims description 14
- 238000012937 correction Methods 0.000 claims description 11
- 238000001514 detection method Methods 0.000 claims description 8
- 101000606504 Drosophila melanogaster Tyrosine-protein kinase-like otk Proteins 0.000 claims description 7
- 230000006978 adaptation Effects 0.000 claims description 7
- 230000007547 defect Effects 0.000 claims description 7
- 230000015556 catabolic process Effects 0.000 claims description 6
- 238000006731 degradation reaction Methods 0.000 claims description 6
- 208000011580 syndromic disease Diseases 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 5
- 230000004044 response Effects 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims 3
- 230000000750 progressive effect Effects 0.000 abstract 1
- 238000007726 management method Methods 0.000 description 29
- 238000013500 data storage Methods 0.000 description 10
- 238000004891 communication Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 238000013473 artificial intelligence Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 5
- 239000007787 solid Substances 0.000 description 4
- 238000007796 conventional method Methods 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 230000003321 amplification Effects 0.000 description 2
- 230000003750 conditioning effect Effects 0.000 description 2
- 239000004744 fabric Substances 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 235000019580 granularity Nutrition 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 239000000696 magnetic material Substances 0.000 description 1
- 239000012092 media component Substances 0.000 description 1
- 238000004377 microelectronic Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 230000005195 poor health Effects 0.000 description 1
- 230000001681 protective effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 229910000679 solder Inorganic materials 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
- 239000010409 thin film Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/18—Error detection or correction; Testing, e.g. of drop-outs
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B5/00—Recording by magnetisation or demagnetisation of a record carrier; Reproducing by magnetic means; Record carriers therefor
- G11B5/455—Arrangements for functional testing of heads; Measuring arrangements for heads
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B27/00—Editing; Indexing; Addressing; Timing or synchronising; Monitoring; Measuring tape travel
- G11B27/36—Monitoring, i.e. supervising the progress of recording or reproducing
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B5/00—Recording by magnetisation or demagnetisation of a record carrier; Reproducing by magnetic means; Record carriers therefor
- G11B5/62—Record carriers characterised by the selection of the material
- G11B5/64—Record carriers characterised by the selection of the material comprising only the magnetic material without bonding agent
- G11B5/66—Record carriers characterised by the selection of the material comprising only the magnetic material without bonding agent the record carriers consisting of several layers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/08—Error detection or correction by redundancy in data representation, e.g. by using checking codes
- G06F11/10—Adding special bits or symbols to the coded information, e.g. parity check, casting out 9's or 11's
- G06F11/1008—Adding special bits or symbols to the coded information, e.g. parity check, casting out 9's or 11's in individual solid state devices
- G06F11/1044—Adding special bits or symbols to the coded information, e.g. parity check, casting out 9's or 11's in individual solid state devices with specific ECC/EDC distribution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/22—Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing
- G06F11/2263—Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
- G06F3/0614—Improving the reliability of storage systems
- G06F3/0616—Improving the reliability of storage systems in relation to life time, e.g. increasing Mean Time Between Failures [MTBF]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
- G06F3/0614—Improving the reliability of storage systems
- G06F3/0619—Improving the reliability of storage systems in relation to data integrity, e.g. data losses, bit errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
- G06F3/0653—Monitoring storage devices or systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0668—Interfaces specially adapted for storage systems adopting a particular infrastructure
- G06F3/0671—In-line storage system
- G06F3/0673—Single storage device
- G06F3/0674—Disk device
- G06F3/0676—Magnetic disk device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/10009—Improvement or modification of read or write signals
- G11B20/10305—Improvement or modification of read or write signals signal quality assessment
- G11B20/10472—Improvement or modification of read or write signals signal quality assessment derived from statistics of other quality measures, e.g. their mean, variance or skew
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/10009—Improvement or modification of read or write signals
- G11B20/10481—Improvement or modification of read or write signals optimisation methods
- G11B20/10518—Improvement or modification of read or write signals optimisation methods using neural networks
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B5/00—Recording by magnetisation or demagnetisation of a record carrier; Reproducing by magnetic means; Record carriers therefor
- G11B5/012—Recording on, or reproducing or erasing from, magnetic disks
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/18—Error detection or correction; Testing, e.g. of drop-outs
- G11B2020/1869—Preventing ageing phenomena from causing data loss, e.g. by monitoring the age of record carriers or by recognising wear, and by copying information elsewhere when a record carrier becomes unreliable
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Computer Security & Cryptography (AREA)
- Quality & Reliability (AREA)
- Computer Hardware Design (AREA)
- Probability & Statistics with Applications (AREA)
- Signal Processing For Digital Recording And Reproducing (AREA)
- Debugging And Monitoring (AREA)
- Digital Magnetic Recording (AREA)
Abstract
Aspects of health management for magnetic storage media are described. In some aspects, the media health manager utilizes a read channel to determine a read metric for a sector of the magnetic storage medium residing in an area of the magnetic storage medium. The media health manager accesses the read metrics of the regions and updates the read metrics of the regions based on the read metrics determined for the sectors to provide updated read metrics for the regions of the magnetic storage medium. The health score for the region of the magnetic storage medium is then determined using a neural network based on the updated read metrics for the region of the magnetic storage medium. By doing so, progressive wear of the magnetic storage medium may be predicted using the health score, thereby enabling replacement of the magnetic storage medium device before the reliability or availability of data stored to the device cannot be improved.
Description
Cross Reference to Related Applications
The present application claims priority from U.S. provisional patent application No.62/799,608 filed on 1 month 31 of 2019 and priority from U.S. patent application No.16/737,174 filed on 8 month 1 of 2020, the disclosures of both of which are incorporated herein by reference in their entireties.
Background
Electronic devices offer many services to modern society. These services enable electronic devices to provide entertainment, aid in scientific research and development, and provide many modern conveniences. Many of these services create or use data stored by the electronic device. The data may include digital media such as books or movies, algorithms to perform complex simulations, personal user data, applications, and the like. To avoid exceeding data storage limits, it is beneficial to increase the data storage capacity of the electronic device and avoid deleting data, limiting services, or purchasing additional external storage devices.
Many electronic devices use media drives to store data on magnetic disks, such as Hard Disk Drives (HDDs). In general, when a disk rotates at a high speed, data stored to an HDD is accessed by moving a read element or a write element on the surface of a medium disk. Thus, HDDs are complex systems composed of interconnected mechanical, electrical, and magnetic components that cooperate to facilitate storage and access of data. Beyond catastrophic events (such as a drop), these mechanical, electrical, and magnetic components of the HDD can wear out gradually over the operational life of the HDD until a catastrophic failure of one of the components prevents data access. If it is not known how much wear has occurred over time, the HDD may fail accidentally before replacement or data backup, resulting in loss of data or reduced uptime of the data storage service.
Disclosure of Invention
This summary is provided to introduce a selection of subject matter that is further described in the detailed description and drawings. Thus, this summary should not be considered to describe essential features, nor should it be used to limit the scope of the claimed subject matter.
In some aspects, a media health manager of a magnetic storage medium implements a method that receives a request to read a sector of the magnetic storage medium that resides in an area of the magnetic storage medium. The method utilizes a read channel to determine a read metric for a sector of the magnetic storage medium that was read based on the request. The method comprises the following steps: the read metrics of the regions of the magnetic storage medium are accessed and updated based on the read metrics determined for the sectors to provide updated read metrics for the regions of the magnetic storage medium. The method then utilizes the neural network to determine a health score for the region of the magnetic storage medium based on the updated read metrics for the region of the magnetic storage medium.
In other aspects, an apparatus comprises: an interface for communicating data with a host; a magnetic disk of a magnetic storage medium for storing data; a read channel to read data from a sector of the magnetic storage medium; and a neural network to determine a media health score. The apparatus also includes a media health manager configured to determine a read metric for a sector of the magnetic storage medium from which data is read using the read channel. The media health manager may also determine an area of the magnetic storage medium in which the sector resides from among a plurality of areas of the magnetic storage medium. Each region of the magnetic storage medium may correspond to a respective subset of tracks on a disk of the magnetic storage medium. The media health manager accesses the read metrics for the regions of the magnetic storage medium and updates the read metrics for the regions based on the read metrics determined for the sectors to provide updated read metrics for the regions of the magnetic storage medium. The media health manager then utilizes the neural network to determine a health score for the region of the magnetic storage medium based on the updated read metrics of the region of the magnetic storage medium.
In other aspects, a system on a chip (SoC) is described that includes a preamplifier interface to receive a signal corresponding to data read from a sector of a magnetic storage medium. The SoC also includes read channel logic to process signals corresponding to data read from the sectors of the magnetic storage medium; a hardware-based processor; and a memory storing processor-executable instructions to implement the media health manager. The media health manager utilizes read channel logic to determine a read metric for a sector of the magnetic storage medium from which data is read. The media health manager may also determine an area of the magnetic storage medium in which the sector resides from among a plurality of areas of the magnetic storage medium. Each region of the magnetic storage medium may correspond to a respective subset of tracks on a disk of the magnetic storage medium. The media health manager accesses the read metrics for the regions of the magnetic storage medium and updates the read metrics for the regions based on the read metrics determined for the sectors to provide updated read metrics for the regions of the magnetic storage medium. The media health manager then utilizes the neural network to determine a health score for the region of the magnetic storage medium based on the updated read metrics of the region of the magnetic storage medium.
The details of one or more implementations are set forth in the accompanying drawings and the description below. Other features and advantages will be apparent from the description and drawings, and from the claims.
Drawings
The details of one or more implementations for health management of magnetic storage media are set forth in the accompanying drawings and the description below. In the drawings, the leftmost digit(s) of a reference number identifies the drawing in which the reference number first appears. The use of the same reference symbols in different instances in the description and the accompanying drawings indicates similar elements:
FIG. 1 illustrates an example operating environment having a device with a magnetic storage medium implemented therein, in accordance with one or more aspects.
Fig. 2 shows an example configuration of the hard disk drive shown in fig. 1.
FIG. 3 illustrates an example configuration of a read channel and a preamplifier according to one or more aspects for health management of a magnetic storage medium.
FIG. 4 illustrates an example configuration of a neural network that may be implemented to calculate a health score for a region of a magnetic storage medium.
Fig. 5 illustrates an example configuration of a neural network that may be implemented to calculate failure probabilities for sectors of a magnetic storage medium.
FIG. 6 depicts an example method for determining a health score for a region of a magnetic storage medium.
FIG. 7 depicts an example method for determining a health score for an area based on an average of read metrics provided by a read channel.
FIG. 8 depicts an example method for determining a storage medium drive health score in accordance with one or more aspects.
FIG. 9 depicts an example method for determining a probability of failure for a sector of a magnetic storage medium.
FIG. 10 illustrates an example system-on-a-chip (SoC) environment for implementing aspects of health management for magnetic storage media.
FIG. 11 illustrates an example storage medium controller configured to implement aspects of health management in a storage medium drive associated with the controller.
Detailed Description
Conventional techniques for tracking Hard Disk Drive (HDD) wear or life expectancy typically only provide a high-level view of actual drive usage, such as operating hours or the number of relocated sectors in the drive. As previously mentioned, HDDs are complex systems composed of interconnected mechanical, electrical, and magnetic components that cooperate to facilitate storage and access of data. With the release of HDDs of each generation, this complexity and tolerance of drive components becomes more important as the write density and storage capacity of the drive increases. Beyond catastrophic events (such as a drop), these mechanical, electrical, and magnetic components of the HDD can wear out gradually over the operational life of the HDD until critical component failures occur and the HDD is scrapped. Loss of HDDs can be very expensive in terms of data loss, data recovery costs, service downtime (e.g., cloud resources), drive replacement costs, and the like. With advanced views of drive usage provided by conventional techniques, users and other storage clients (e.g., data centers) are often unable to accurately estimate when an HDD will fail during its intended lifetime. Therefore, many HDDs fail unexpectedly before drive replacement or data backup is achieved, resulting in loss of stored data or reduced availability of data storage services.
The present disclosure describes apparatus and techniques for health management of magnetic storage media. In contrast to conventional magnet writing techniques, the described apparatus and techniques use metrics provided by the read channel that can more accurately capture the gradual wear or degradation of components of the media drive directly or indirectly. Typically, the read channel of a storage medium drive processes and decodes signals as data is read from the storage medium disk. Signal processing and decoding of the read signal by the read channel may include signal conditioning, sampling, equalization, detection, error correction, and the like. In this way, the read channel may determine or generate various metrics (e.g., low level simulation and error correction metrics) that reflect the respective conditions of the electrical, mechanical, or magnetic components of the storage medium drive.
Based on the metrics provided by the read channel, the media health manager may utilize the neural network to determine or update a health score for a portion or region of the magnetic storage medium. In some aspects, the neural network is pre-trained to classify a large number of storage medium metrics into several or several health scores, which is useful for determining the health of the magnetic storage medium or the health of the storage medium drive. By doing so, the internal operation of the storage medium drive may be optimized to avoid areas of the magnetic storage medium that are poor in health, or the storage medium drive may be replaced based on the health prediction. For example, most data centers or cloud service providers run millions of HDDs that may be deployed in one or more locations. With the health score provided by aspects of health management for magnetic storage media, a data center administrator may effectively implement HDD replacement or access redirection (e.g., to another HDD) before HDD failure. This may enable an administrator to protect mission critical data and better ensure 24/7 availability of the data server HDD to comply with various Service Level Agreements (SLAs) for data access. In other cases, the media health manager may determine a probability of failure for one or more sectors of the magnetic storage medium. When a read sector may fail, the media health manager may implement more aggressive error correction, or inform the service provider (e.g., cloud) that the data may not be available, thereby enabling the provider to obtain the required data from another drive. In some cases, this may effectively enable the service provider to meet the conditions of the SLA by providing the required data from the alternative location before waiting until the final read failure of the low probability sector occurs.
In various aspects for health management of magnetic storage media, a media health manager may utilize a read channel to determine a read metric for a sector of the magnetic storage media residing in an area of the magnetic storage media. The media health manager accesses the read metrics of the regions and updates the read metrics of the regions based on the read metrics determined for the sectors to provide updated read metrics for the regions of the magnetic storage medium. The media health manager then determines a health score for the region of the magnetic storage medium based on the updated read metrics for the region of the magnetic storage medium using the neural network. By doing so, the health score may be used to predict gradual wear of the magnetic storage medium, thereby enabling replacement of the magnetic storage medium device prior to failure to improve reliability or availability of data stored to the magnetic storage medium device.
The following discussion describes an operating environment, techniques that may be employed in the operating environment, and a system on a chip (SoC) in which components of the operating environment may be implemented. In the context of the present disclosure, reference is made to an operating environment by way of example only.
Operating environment
FIG. 1 illustrates an example operating environment 100 having a computing device 102, the computing device 102 being capable of storing or accessing various forms of data or information. Examples of computing device 102 may include a laptop 104, a desktop 106, and a server 108, any of which may be configured to store a portion of a network or cloud storage. Other examples of computing device 102 (not shown) may include a tablet computer, a set-top box, a data storage device, a wearable smart device, a television, a content streaming device, a High Definition Multimedia Interface (HDMI) media stick, a smart appliance, a home automation controller, a smart thermostat, an internet of things (IoT) device, a Mobile Internet Device (MID), a Network Attached Storage (NAS) drive, an aggregate storage system, a gaming machine, an automobile entertainment device, an automobile computing system, an automobile control module (e.g., an engine or powertrain control module), and so forth.
In general, computing device 102 may provide, transfer, or store data for any suitable purpose, such as enabling functionality of a particular type of device, providing a user interface, enabling network access, implementing gaming applications, playing back media, providing navigation, editing content, providing data storage, and so forth. Alternatively or additionally, the computing device 102 can store various data, such as databases, user data, multimedia, applications, operating systems, and the like. One or more computing devices 102 may be configured to provide remote data storage or services, such as cloud storage, archiving, backup, client services, record keeping, and the like.
The computing device 102 includes a processor 110 and a computer readable storage medium (CRM) 112. Processor 110 may be implemented as any suitable type or number of processors, single or multi-core (e.g., ARM or x86 processor cores), for executing instructions or commands of an operating system or other program of computing device 102. CRM 112 includes a storage medium 114 and a medium drive 116. The memory medium or system memory of computing device 102 may include any suitable type or combination of volatile memory or non-volatile memory. For example, the volatile memory of computing device 102 may include various types of Random Access Memory (RAM), dynamic RAM (DRAM), static RAM (SRAM), and the like. The non-volatile memory may include Read Only Memory (ROM), electrically Erasable Programmable ROM (EEPROM), or flash memory (e.g., NOR flash memory or NAND flash memory). These memories may store data associated with the applications and/or operating systems of computing device 102, alone or in combination.
The media drive 116 of the computing device 102 may comprise one or more media drives or may be implemented as part of a data storage system associated with the computing device 102. In this example, the media drive 116 includes a Hard Disk Drive (HDD) 118, which hard disk drive 118 is capable of storing data and is described with reference to various aspects for health management of magnetic storage media. Alternatively or additionally, the media drive 116 may be configured as any suitable type of data storage drive or system, such as a storage device, storage drive, storage array, storage volume, or the like. Although described with reference to computing device 102, media drive 116 may also be implemented alone as a standalone device or as part of a larger storage collection, such as a data center, server farm, or virtualized storage system (e.g., for cloud-based storage or services), in which aspects for health management of magnetic storage media are implemented.
Computing device 102 may also include I/O ports 120, a graphics processing unit (GPU, not shown), and a data interface 122. In general, the I/O ports 120 allow the computing device 102 to interact with other devices, peripherals, or users. For example, the I/O ports 120 may include or be coupled to a universal serial bus, a human interface device, an audio input, an audio output, and the like. The GPU processes and renders graphics-related data of the computing device 102, such as user interface elements of an operating system, applications, and the like. In some cases, the GPU accesses a portion of local memory to render graphics, or the GPU includes dedicated memory for rendering graphics (e.g., video RAM) of the computing device 102.
The data interface 122 of the computing device 102 provides connectivity to one or more networks and other devices connected to those networks. The data interface 122 may include a wired interface, such as an ethernet interface or a fiber optic interface for data transmitted over a local area network, an intranet, or the internet. Alternatively or additionally, data interface 122 may include wireless interfaces that facilitate communication over a wireless network, such as a Wireless Local Area Network (WLAN), a wide area wireless network (WAN, e.g., a cellular network), and/or a Wireless Personal Area Network (WPAN). Any data transferred through the I/O port 120 or data interface 122 may be written to or read from the storage system of the computing device 102 in accordance with one or more aspects for health management of the magnetic storage medium.
Returning to the media drive 116, the computing device 102 may include a hard disk drive 118 as shown and/or other types of storage media for which aspects of health management may be implemented. Although not shown, other configurations of media drive 116 are also contemplated, such as Solid State Drives (SSDs), tape drives, optical media drives, HDD/SSD hybrid drives, and other storage systems that write data to a memory medium (e.g., a magnetic or optical storage medium). Alternatively or additionally, the computing device 102 may include an array of media drives, or serve as a media drive aggregation device or host for multiple media drives in which aspects of health management may be implemented.
In this example, the disk drive 118 includes a head-disk assembly (HDA) 124 and a drive control module 126 to implement or enable the functions of the hard disk drive 118. In some cases, the driver control module 126 is implemented as a Printed Circuit Board Assembly (PCBA) with semiconductor devices, logic, or other circuitry. The HDA 124 includes one or more media disks 128 (or platters) mounted on an integrated spindle and motor (e.g., voice Coil Motor (VCM)) assembly 130. The spindle and motor assembly 130 may rotate the media disk 128 under (or over) a read/write head 132 coupled to a head assembly (not shown) of the HAD 124. The media disk 128 may be coated with a hard magnetic material (e.g., a granular surface or a thin film surface) and may be written to or read from one or both sides. The read/write head 132 may be operably coupled to a preamplifier/writer module 134 (pre-amp/writer 134) of the HDA 124, the preamplifier/writer module 134 including preamplifier circuitry and other logic for amplifying the write signal and the read signal, respectively. The preamplifier/writer 134 may receive or store a selection, amplification, or sense current value of a head useful for writing data to the magnetic medium 202 or reading data from the magnetic medium 202.
As shown in FIG. 1, the example drive control module 126 of the hard disk drive 118 may include a storage medium controller 136, a servo control unit 138, and a read/write channel 140 (read channel 140). The storage medium controller 136 enables the computing device 102 to access the contents of the magnetic storage medium of the media drive 116, such as an operating system, applications, or data for applications or other services. The storage medium controller may also write data of the computing device 102 to and read data of the computing device 102 from the magnetic storage medium of the media drive. In general, the drive control module 126 may direct or use the servo control unit 138 to control mechanical operations, such as positioning of the read/write head 132 through the HDA 124 and rotational speed control through the spindle and motor assembly 130.
The read/write channel 140 may include a digital-to-analog path and an analog-to-digital path for converting write data to a write signal or a read signal to read data, respectively. For example, read channel 140 may process and decode signals when data is read from media disk 128. This signal processing and decoding of the read signal may include signal conditioning, sampling, equalization, detection, error correction, and the like. Throughout this disclosure, the implementation and use of the read channel 140 is varied and described. The drive control module 126, or components thereof, may be implemented as one or more IC chips, a system-on-a-chip, a system-in-package, or a microprocessor provided with or implementing a hard disk drive controller. The drive control module 126 may also include drive electronics (not shown) and/or include various interfaces, such as a host bus interface, a storage medium interface, a spindle interface, or a preamplifier/writer interface.
In some aspects, the read channel 140 includes or is associated with a media health manager 142, a neural network 144, and media health data 146. The media health manager 142 may obtain metrics from the read channel 140 for sectors of the magnetic storage medium from which data is read. Based on the metrics provided by the read channel 140, the media health manager 142 may determine or update a health score for a portion or region of the magnetic storage medium, such as by using the neural network 144. One or more of these neural networks 144 may be pre-trained to classify a large number of storage medium metrics into several or several health scores, which is useful for determining the health of the magnetic storage medium or the health of the storage medium drive. The media health data 146 may include previously calculated health scores for regions or portions of the magnetic storage medium, as well as an average of read channel metrics for sectors or tracks of the magnetic storage medium. In some cases, the media health manager 142 stores the health score determined for a portion or area of the magnetic storage medium to the media health data 146, which media health data 146 may be part of an information table for the portion or area of the magnetic storage medium. Throughout this disclosure, the implementation and use of the media health manager 142, the neural network 144, or the media health data 146 are varied and described.
By way of example, consider fig. 2 (not shown in the protective enclosure), shown generally at 200, which provides an example configuration of hard disk drive 118. As shown in FIG. 2, the HDA 124 of the hard disk drive 118 includes an integrated spindle and motor assembly 130, through which the media disk 128 (or platter) of the magnetic storage medium 202 is supported and/or operated. In general, each media disk 128 may be divided or partitioned into predetermined areas or regions of magnetic storage medium 202. In this example, the surface 204 of the media disk 128 is divided into or includes three regions 206 (e.g., circular bands/rings or wedges) that include respective subsets of tracks 208 and sectors 210 of the magnetic storage medium 202. Other media disks of media disks 128, or the surfaces of those disks, may be configured similar to or different from those shown in fig. 2 (e.g., a different number of regions). In some aspects, the media health manager maintains drive health in the form of an area health score card for one or more areas 206 of the magnetic storage medium.
During operation, the arm 212 may maneuver and thus position the read/write head 132 (or heads 132) over a desired track 208 or sector 210 of the magnetic storage medium 202 on the media disk 128. In various aspects, the read/write head 132 may include a variety of numbers of head elements having combined or separate functions (e.g., dedicated R/W functions). For example, the read/write head 132 may include one or more readers (read heads/elements) and one writer (write heads/elements). In other cases, the read/write head 132 may include a dedicated write head (element) and one or more separate additional dedicated read heads (elements). Alternatively or additionally, although multiple arms 212 are shown in fig. 2, the HDA 124 or spindle and motor assembly may be implemented with a single arm 212 or other suitable structure for positioning the read/write head 132.
The HDA 124 and the drive control module 126 may be implemented separately, on separate substrates, and/or as separate PCBAs of media drives. The signals or data communicated between the HDA 124 and the drive control module 126 may be carried by flexible printed cables or other suitable connection structures such as tracks, connectors, bond wires, solder balls, etc. The HDA 124 of the hard disk drive 118 may be configured to perform write operations according to any suitable recording technique, such as Perpendicular Magnetic Recording (PMR), laminated magnetic recording (SMR), thermally assisted magnetic recording (HAMR), microwave Assisted Magnetic Recording (MAMR), and the like.
FIG. 2 also illustrates, at 214, a plan view of the media disk 128 having the region 206 of the magnetic storage medium 202 implemented in accordance with one or more aspects. In this example, tracks 208 and sectors 210 of surface 204 of disk 216 are organized or divided into three areas of magnetic storage media, including area 1 206-1, area 2 206-2, and area 3 206-3. Although illustrated with three regions 206, the platters 216 or surface 204 of the media disks 128 may be organized or include any suitable number of regions 206, such as regions 1-1 through n206-n, where n is any suitable number. In general, each region 206 may be configured as a subset of tracks 208 and sectors 210 of the magnetic storage medium 202 that include the surface 204. In some cases, the regions 206 are configured such that each region includes approximately equal numbers of sectors 210 or surface areas of the platters 216.
In various aspects of health management of magnetic storage media, the media health manager 142 may maintain health scores for the regions 206 of the media disks 128. Generally, when accessing the sector 210 of the region 206 to read data from the sector 210 of the magnetic storage medium, the region health score may be updated based on the metrics provided by the read channel 140. The health score for a region of the magnetic storage medium may be indicative of a wear level, a degradation level, or a reliability level of the region of the magnetic storage medium. Alternatively or additionally, these zone health scores may also be compiled for multiple media disks 128 of the hard disk drive 118 to enable determination of a drive health score for the hard disk drive 118.
FIG. 3 illustrates an example configuration of a read channel and a preamplifier, generally at 300, implemented in accordance with one or more aspects for health management of a magnetic storage medium. In this example, the media health manager 142 and the analog front end 302 are operably coupled with the read channel 140. In addition, the neural network 144 and the media health data 146 are operatively coupled to or accessible by the media health manager 142 of the read channel 140. Although shown as separate components or circuitry in fig. 3, the read channel 140 and the media health manager 142 may be integrated as one component separate between other components of the hard disk drive 118 and/or with other microelectronic devices or circuitry of the preamplifier 134 and/or the read channel 140.
In this example, the read channel 140 and other components are described in the context of reading data from a sector of the magnetic storage medium 202 (magnetic medium 202). For example, the host system or computing device 102 may issue a read command for data stored to one or more sectors 210 of the media disk 128. As the media disk 128 rotates under the read head 132, the read head 132 senses the magnetic field 304 of the data stored to the magnetic media 202, which generates an analog signal 306 at the read head 132. The pre-amplifier 134 (pre-amp 134) amplifies the analog signal 306 received from the read head 132 and provides an amplified signal 308 to the analog front end 302 of the read channel 140.
Typically, the analog front end 302 conditions and samples an amplified signal 308 (e.g., a read back continuous time signal) provided by the pre-amplifier 134. The read channel 140 converts the sampled signal to a digital signal and recovers the decoded data 310, which decoded data 310 is provided to the storage medium controller 136. The read channel 140 may include any suitable combination of equalizer modules, detector modules, adaptation modules, or gain modules for detecting, equalizing, and/or decoding the data 310 from the signals received from the pre-amplifier 134. Ending this example, the storage controller 136 may then send the decoded data 310 as read data 312 to the host interface 314 to complete the read command issued by the host system or computing device 102.
With respect to various data access operations, such as read operations for data stored to one or more sectors 210 of the media disk 128, the media health manager 142 may implement various aspects of health management for the magnetic storage media. In general, the zone information table and/or media health data may include calibration parameters associated with or available to access a particular zone of the magnetic storage medium. The media health manager 142 or read channel 140 may maintain the status of the magnetic storage media health in the form of a zone health score or score card, such as by ranking the health of the zone from 1 (e.g., bad) to 5 (e.g., excellent). The health score for an area may be updated or recalculated each time a sector is read from that particular area. In certain aspects, the media health manager generates or updates the health score by obtaining various internal metrics from the read channels of a given sector of the area (e.g., during a read operation), and determines the health score based on the internal metrics using a machine learning algorithm or neural network. These internal metrics may include unrecovered sectors, retry method counts, average errors corrected by error correction codes, log Likelihood Ratio (LLR) Probability Density Function (PDF) statistics, sync mark errors, various cyclic errors, and the like.
For example, when the media disk 128 is looking for a particular region 206 to read one or more sectors 210, the media health manager 142 may read previously calculated region metrics from the media health data 146, which media health data 146 may be implemented as part of a region information table (not shown). Based on the current metrics for the sectors 210 of the area 206 being read, the media health manager updates the long-term and short-term averages of the metrics for those sectors 210. The media health manager may then determine a health score for the region 206 based on the updated long-term and short-term metrics using the neural network. The neural network may be pre-trained to map a wide variety of read channel metrics to several or several health status states. The updated health score and updated metrics for the region 206 may also be written back to the media health data 146 or region information table upon exiting a given region.
Various aspects of health management for magnetic storage media described throughout this disclosure may be implemented by the media health manager 142, the media health manager 142 interacting with a neural network 144 (e.g., fig. 4 and 5) or a storage system controller or read channel, or any suitable Artificial Intelligence (AI) engine, AI model, or AI driver associated with a storage system controller or read channel. Regarding processing various metrics (e.g., read metrics, signal metrics, or electromechanical metrics) of a storage medium drive or system, one or more neural networks 144 may be implemented using machine learning based on one or more neural networks (e.g., pre-trained) to make medium health scores, component wear estimates, or predict drive health. Any AI model, AI algorithm, or neural network of the media health manager 142 may include a set of connected nodes, such as neurons or perceptrons, organized into one or more layers.
An example of the neural network 144 associated with the media health manager 142 may be implemented with a deep neural network that includes an input layer, an output layer, and one or more hidden middle layers located between the input layer and the output layer of the neural network. Each node of the deep neural network may in turn be fully connected or partially connected between the layers of the neural network. The neural network 144 may be any Deep Neural Network (DNN), such as a Convolutional Neural Network (CNN) including one of AlexNet, resNet, ***Net, mobileNet, etc. Alternatively or additionally, the neural network 144 may include any suitable Recurrent Neural Network (RNN) or any variation thereof. In general, the AI model or neural network employed by the media health manager 142 may also include any other supervised learning, unsupervised learning, reinforcement learning algorithms, and the like.
In various aspects, the neural network 144 may be implemented as a recurrent neural network in which the connections between nodes form loops to retain information from a previous portion of the input data sequence for a subsequent portion of the input data sequence (e.g., to read internal metrics of the channel). Alternatively, the neural network 144 may be implemented as a feed-forward neural network having connections between nodes that do not form loops between input data sequences. In other cases, the neural network 144 of the media health manager 142 may include a Convolutional Neural Network (CNN) with a multi-layer perceptron, where each neuron in a given layer is connected with all neurons of an adjacent layer. In certain aspects, the neural network 144 is based on a convolutional neural network that may be applied to previous media health scores to predict or forecast some form of subsequent or future health trend of the magnetic storage medium. Alternatively or additionally, the neural network 144 may include or utilize various regression models, such as multiple linear regression models, a single linear regression model, a logistic regression model, a stepwise regression model, a multivariate adaptive regression model, a locally estimated scatter model, and the like.
By way of example, consider fig. 4 in which an example configuration of a neural network is shown generally at 400. The example neural network 144 is implemented in accordance with one or more aspects to calculate a health score for a region of a magnetic storage medium. In general, the neural network 144 is configured to map or classify the internal metrics 402 (read metrics 402) of the read channel 140 to the region health score 404 for a region of the magnetic storage medium.
As shown in fig. 4, the input to the neural network 144 may include any suitable number of internal metrics 402 of the read channel 140. In certain aspects, the neural network 144 is pre-trained to map the read metrics 402 to one of five region health scores 404 ranging from bad, alert, fair, good, and excellent. The read channel 140 may generate or have access to a wide variety of internal metrics, which may be useful for calculating or determining health scores for regions of the magnetic storage medium. In this example, the read metrics include unrecovered sector rate 406, average retry method rate 408, average of off-track detection 410, average number of seek errors 412, average number of errors corrected by Error Correction Codes (ECCs) 414, average syndrome (syndrome) weights in ECCs 416, average number of ECC iterations 418, average of Log Likelihood Ratio (LLR) Probability Density Function (PDF) statistics 420, average Mean Square Error (MSE) of the front-end loop signal 422, average length of detected media defects 424, average length of analog-to-digital converter (ADC) saturation 426, average delta of adaptation parameters 428, average of Sync Mark (SM) errors 430, or other various metrics 432. In this way, based on detailed internal metrics available to the media health manager 142, the neural network 144 may determine an area health score that predicts storage media drive wear or reliability with greater accuracy than conventional techniques that track drive life through hour or sector remapping.
FIG. 5 illustrates, at 500, an example configuration of a neural network to calculate failure probabilities for sectors of a magnetic storage medium, implemented in accordance with one or more aspects. In some aspects, the probability of a sector failure may be useful for optimizing queue management of data or selecting a particular pipeline for decoding data for a sector. For example, the front-end portion of the read channel may classify each sector as one of three or five signal-to-noise levels, which classification may be used by the back-end portion of the read channel to select a pipeline for decoding each sector to improve decoding efficiency. In this example, the neural network 144 is configured to determine a probability 504 of sector failure for the sector based on the front end metrics 502. Using the probability 504 of the sector failing, the back end of the read channel 140 may select a pipeline for decoding the sector that is more likely to succeed by default by stepping through multiple pipelines than because the initial attempt to decode the sector failed. Alternatively or additionally, the media health manager may inform the service provider (e.g., cloud) that the data is likely not available from a low probability sector, thereby enabling the provider to obtain the required data from another drive without waiting until the read operation eventually fails.
As shown in fig. 5, the front end of the read channel 140 may generate or have access to a variety of internal metrics, which may be useful for calculating or determining the probability of a sector failure. In this example, the front end metrics include an indication of: syncmark discovery 506, syncmark distribution 508, detected off-track 510, incremental feedforward finite impulse response (FAFIR) filter tap 512, incremental FIR-3T tap 514, baseline accumulator 516, frequency accumulator 518, mutual information 520 from Vmm and non-return to zero (NRZ), vmm counts 522 with various thresholds, MSE 524 calculated from error signals, length 526 of detected defect markers, incremental gain variation 528, incremental ASC variation 530, or other various metrics 532. In this way, based on detailed internal metrics available to the read channel 140, the neural network 144 can predict the probability 504 of sector failure, which is useful for altering or optimizing pipeline selection to improve decoding of non-optimal sectors.
Techniques for health management of magnetic storage media
The following discussion describes techniques for health management of magnetic storage media that enable improving reliability or availability (e.g., uptime) of data by predicting or tracking health (e.g., wear) of magnetic storage media components. These techniques may be implemented using any of the environments and entities described herein, such as a read channel 140, a media health manager 142, a neural network 144, or media health data 146. These techniques include methods 600 through 900 shown in fig. 6 through 9, each of which is shown as a set of operations performed by one or more entities.
The methods are not necessarily limited to the order of operations shown in the associated drawings. Rather, any operations may be repeated, skipped, replaced, or reordered to implement aspects described herein. Furthermore, the methods may be used in whole or in part in combination with one another, whether performed by the same entity, separate entities, or any combination thereof. For example, aspects of the described methods may be combined to achieve health scores for various granularities of magnetic storage media, such as corresponding health scores for areas, surfaces, disks, or drives of the magnetic storage media. In portions of the following discussion, reference will be made to the operating environment 100 of FIG. 1, the entities of FIGS. 2-5. Such references should not be considered as limiting the described aspects to operating environment 100, entities, configurations, or implementations, but rather as illustrations of one of various examples. Alternatively or additionally, the operations of the method may also be implemented by or with the entities described with reference to the system on chip of fig. 10 and/or the storage medium controller of fig. 11.
FIG. 6 depicts an example method 600 for determining a health score for an area of a magnetic storage medium, including operations performed by or utilizing a read channel 140, a medium health manager 142, a neural network 144, and/or medium health data 146.
At 602, a request to read a sector of a magnetic storage medium is received. A sector of a magnetic storage medium may reside in an area of the magnetic storage medium. In some cases, a read channel receives a request from a storage medium controller to read a sector (or data from a sector). The request may be one of a plurality of requests that cause a plurality of corresponding sectors to be read from a particular sector of a media disk or platter of a magnetic storage medium. Alternatively or additionally, the media health manager may determine the region of the magnetic storage medium in which the sector resides from among a plurality of regions of the magnetic storage medium.
At 604, a read metric is determined for a sector having a read channel. A read metric is determined for the sectors that were read based on the request. In some cases, the read metrics include internal metrics of the read channel, such as unrecovered sector rate, average retry method rate, average off-track detection, average number of seek errors, average number of errors corrected by Error Correction Codes (ECCs), average syndrome weights in ECCs, average number of ECC iterations, average of Log Likelihood Ratio (LLR) Probability Density Function (PDF) statistics, average Mean Square Error (MSE) of front-end loop signals, average length of detected media defects, average length of analog-to-digital converter (ADC) saturation, average delta of adaptation parameters, or average of sync mark errors.
At 606, read metrics of an area of the magnetic storage medium are accessed. These read metrics may include previously determined or calculated read metrics for the sectors of the region. In some cases, an area information table or media health data store is accessed to obtain a read metric for an area. In general, the read metrics for the region may include corresponding read metrics for the sector being read and other sectors residing in the region of the magnetic storage medium. Alternatively or additionally, the read metrics of a sector may include a long-term average of a particular read metric and a short-term average of a particular read metric.
At 608, the read metrics of the regions of the magnetic storage medium are updated based on the read metrics determined for the sectors of the magnetic storage medium. The read metric for each sector read from the region may be updated based on the current corresponding read metric provided by the read channel for that sector. In some cases, updating may include updating both the long-term average and the short-term average of the metrics for the sector.
At 610, a health score is determined for a region of the magnetic storage medium using the neural network. The neural network determines a health score based on the updated read metrics of the region of the magnetic storage medium. The neural network may be configured or pre-trained to map the updated read metrics of the read channel to one of at least three health score classifications to determine a health score for a region of the magnetic storage medium. The health score for a region of the magnetic storage medium may be indicative of a wear level, a degradation level, or a reliability level of the region of the magnetic storage medium.
Optionally, at 612, a health score for a region of the magnetic storage medium is stored. The health score for a region of the magnetic storage medium may be written to a region information table or medium health data store. Alternatively or additionally, updated read metrics for sectors of an area may also be written back into the area information table or medium health data store.
Optionally, at 614, the health score of the region of the magnetic storage medium is transmitted to the storage medium controller. In some cases, a health score for an area of the magnetic storage medium is sent to the storage medium controller such that the storage medium controller can compile the health score with other metrics to determine a health score for a media drive or higher level storage entity associated with the storage medium controller.
FIG. 7 depicts an example method 700 for determining a health score for an area based on an average of read metrics provided by a read channel, including operations performed by or utilizing the read channel 140, the media health manager 142, the neural network 144, and/or the media health data 146.
At 702, a read metric for a sector of a magnetic storage medium is determined using a read channel. The sectors of the magnetic storage medium may reside in or within regions of the magnetic storage medium. In some cases, respective read metrics are determined for a plurality of sectors read from an area of the magnetic storage medium. Alternatively or additionally, the media health manager may determine the region of the magnetic storage medium in which the sector resides from among a plurality of regions of the magnetic storage medium.
At 704, a previously determined read metric for the region is read from the region information table. In general, the previously determined read metrics for an area may include corresponding read metrics for the sector being read and other sectors residing in the area of the magnetic storage medium. Alternatively or additionally, the read metrics of a sector may include a long-term average of a particular read metric and a short-term average of a particular read metric.
Optionally, at 706, a short term average of the read metrics for the region is updated based on the read metrics determined for the sector. Typically, the short term average of the read metrics for a sector is updated based on the current read metrics provided by the read channel. In the case of reading a plurality of sectors from an area, a respective short-term average of the read metrics may be updated for each of the plurality of sectors read from the area.
Optionally, at 708, a long-term average of the read metrics for the region is updated based on the read metrics determined for the sector. Typically, the long-term average of the read metrics for the sectors is updated based on the current read metrics provided by the read channel. In the case of reading a plurality of sectors from an area, a respective long-term average of the read metrics may be updated for each of the plurality of sectors read from the area.
At 710, a health score is determined for a region of the magnetic storage medium using a neural network. The neural network determines a health score based on the updated short-term average and/or long-term average of the read metrics of the region. The neural network may be configured or pre-trained to map the updated short-term average and/or long-term average of the read metrics of the region to one of at least three health score classifications to determine a health score for the region of the magnetic storage medium. The health score for a region of the magnetic storage medium may be indicative of a wear level, a degradation level, or a reliability level of the region of the magnetic storage medium.
At 712, the health score of the region of the magnetic storage medium is stored to the region information table. Alternatively or additionally, the health score for a region of the magnetic storage medium is written to the medium health data store. At 714, updated short-term and long-term averages of the read metrics of the region are stored to the region information table. In some cases, updated short-term and long-term averages of the read metrics of the region may be written to the media health data store.
Optionally, at 716, the health score of the region of the magnetic storage medium is transmitted to the storage medium controller. In some cases, a health score for an area of the magnetic storage medium is sent to the storage medium controller such that the storage medium controller can compile the health score with other metrics (e.g., electromechanical or signal metrics) and use the combined metrics to determine a health score for a medium drive or higher level storage entity associated with the storage medium controller.
FIG. 8 depicts an example method 800 for determining a storage media drive health score, including operations performed by or with the read channel 140, the media health manager 142, the neural network 144, and/or the media health data 146, in accordance with one or more aspects.
At 802, respective health scores for one or more regions of a magnetic storage medium are received from a read channel of a storage medium drive. The respective health score may be received from a media health manager or a read channel associated with a disk or surface of a media disk on which the area is implemented or organized.
At 804, an electromechanical metric of the storage media drive is received from a spindle and voice coil motor assembly of the storage media drive. The electromechanical metric may include any suitable metric related to: parameters, calibration information, or performance data associated with a spindle of a storage medium drive, a voice coil motor, or other electromechanical component.
At 806, signal metrics of the storage medium drive are received from a preamplifier of the storage medium drive. The signal metrics may include various amplification or gain settings of the pre-amplifier. In some cases, the signal metrics include respective signal metrics for a plurality of platters or surfaces of a magnetic storage medium of the storage medium drive.
At 808, a drive health score is determined for the storage medium drive based on the respective health score, the electromechanical metric, and the signal metric for the region. In some cases, at least two of the health score, the electromechanical metric, or the signal metric are used to determine an overall or final health score for the storage medium drive. The health score for a storage medium drive may indicate a wear level, a degradation level, or a reliability level of the storage medium drive.
FIG. 9 depicts an example method 900 for determining a probability of failure for a sector of a magnetic storage medium, including operations performed by or utilizing the analog front end 302, the medium health manager 142, the neural network 144, and/or the medium health data 146.
At 902, a request to read a sector of a magnetic storage medium is received. The request may be received from a storage media controller associated with a media disk on which the sector resides. In some cases, the storage medium controller requests to read the plurality of sectors as part of a read command received from the host system or computing device.
At 904, a signal metric is received from a front end of a read channel for a sector of a magnetic storage medium. The front end of the read channel 140 may generate or have access to a variety of internal metrics, which may be useful in calculating or determining the probability of a sector failure. For example, the front-end metrics may include an indication of: syncmark discovery, syncmark distribution, detected off-track, incremental FAFIR filter taps, incremental FIR-3T taps, baseline accumulator, frequency accumulator, mutual information from Vmm and NRZ, vmm counts with various thresholds, MSE calculated from error signals, length of detected defect markers, incremental gain variation, incremental ASC variation, or other various metrics.
At 906, a probability of failure for the sector is determined using a neural network based on the signal metric for the sector. The neural network 144 may be configured or pre-trained to determine the probability of sector failure based on signal metrics provided by the front end.
At 908, the probability of failure for the sector is provided to the back end of the read channel. The backend may use the probability of sector failure to select decoding parameters for the sectors to improve the backend's decoding, detection, or error correction operations. In some cases, the back end of the read channel may select a pipeline for decoding a sector that is more likely to succeed than stepping through multiple pipelines by default because the initial attempt to decode the sector failed.
System on chip
FIG. 10 illustrates an exemplary system on a chip (SoC) 1000 that can implement aspects of health management for magnetic storage media. SoC 1000 may be implemented in any suitable device, such as a smart phone, a netbook, a tablet, an access point, network attached storage, a camera, a smart appliance, a printer, a set top box, a server, a Solid State Drive (SSD), a tape drive, a Hard Disk Drive (HDD), a storage drive array, a memory module, a storage media controller, a storage media interface, a head-disk assembly, a magnetic media preamplifier, an automotive computing system, or any other suitable type of device (e.g., other devices described herein). Although described with reference to a SoC, the entities of fig. 10 may also be implemented as other types of integrated circuits or embedded systems, such as Application Specific Integrated Circuits (ASICs), memory controllers, read channel components, memory controllers, communication controllers, application Specific Standard Products (ASSPs), digital Signal Processors (DSPs), programmable socs (psos), system in package (sips), or Field Programmable Gate Arrays (FPGAs).
SoC1000 may be integrated with electronic circuitry, microprocessors, memory, input-output (I/O) control logic, communication interfaces, firmware, and/or software that may be used to provide functionality of a computing device or magnetic storage system, such as any of the devices or components described herein (e.g., a hard disk drive). SoC1000 may also include an integrated data bus or interconnect structure (not shown) that couples the various components of the SoC for data communication or routing between the components. The integrated data bus, interconnect fabric, or other components of SoC1000 may be disclosed or accessed through an external port, parallel data interface, serial data interface, peripheral component interface, or any other suitable data interface. For example, components of SoC1000 may access or control external storage media or magnetic read circuitry through external interfaces or off-chip data interfaces.
In this example, soC1000 is shown with various components including input-output (I/O) control logic 1002 and a hardware-based processor 1004 (processor 1004), such as a microprocessor, processor core, application processor, DSP, and the like. SoC1000 also includes memory 1006, which memory 1006 may include any type and/or combination of RAM, SRAM, DRAM, non-volatile memory, ROM, one-time programmable (OTP) memory, multiple-time programmable (MTP) memory, flash memory, and/or other suitable electronic data storage. In certain aspects, the processor 1004 and code (e.g., firmware) stored on the memory 1006 are implemented as a read/write channel module or as part of a storage medium interface to provide various functions associated with health management for magnetic storage media. In the context of the present disclosure, the memory 1006 stores data, code, instructions, or other information via non-transitory signals, and does not include carrier waves or transitory signals. Alternatively or additionally, soC1000 may include a data interface (not shown) for accessing additional or expandable off-chip storage media, such as magnetic memory or solid state memory (e.g., flash memory or NAND memory).
SoC 1000 may also include firmware 1008, applications, programs, software, and/or an operating system, which may be implemented as processor-executable instructions maintained on memory 1006 for execution by processor 1004 to implement the functions of SoC 1000. In this example, in accordance with one or more aspects, soC 1000 includes a preamplifier interface 1010 to receive a signal corresponding to data read from a sector of a magnetic storage medium. SoC 1000 may also include other communication interfaces, such as a transceiver interface for controlling or communicating with components of a local on-chip (not shown) or off-chip communication transceiver. Alternatively or additionally, the transceiver interface may also include or implement a signal interface to transmit Radio Frequency (RF), intermediate Frequency (IF), or baseband frequency signals off-chip to facilitate wired or wireless communication through a transceiver, physical layer transceiver (PHY), or Medium Access Controller (MAC) coupled to the SoC 1000. For example, soC 1000 may include a transceiver interface configured to enable storage over a wired or wireless network, such as providing a Network Attached Storage (NAS) device with media health management features.
SoC 1000 also includes analog front end 302 and read channel logic 1012 (e.g., a front end portion) for processing signals received from the pre-amplifier via pre-amplifier interface 1010. Typically, the analog front end 302 conditions and samples the read signal (e.g., readback continuous time signal) provided by the pre-amplifier. The read channel logic 1012 may include any suitable combination of equalizer modules, detector modules, adaptation modules, or gain modules for detecting, equalizing, and decoding data from the read signal received from the pre-amplifier. In some aspects, soC 1000 includes a media health manager 142, a neural network 144, and media health data 146, which may be implemented separately as shown, or in combination with a processing component or data interface. Alternatively or additionally, soC 1000 may include an interface to a storage medium controller or spindle/motor assembly of a magnetic media disk drive.
As described herein, the media health manager may receive read metrics (e.g., internal read channels or analog front end metrics) from the read channels and determine or update health scores for regions of the magnetic storage medium based on these metrics and/or using the neural network to implement various aspects of health management for the magnetic storage medium. Any of these entities may be implemented as distinct or combined components, as described with reference to the various aspects presented herein. Examples of such components and/or entities, or respective functions, are described with reference to respective components or entities of environment 100 of fig. 1 or respective configurations shown in fig. 2-5. The media health manager 142 may be implemented in whole or in part as digital logic, circuitry, and/or processor-executable instructions maintained by the memory 1006 and executed by the processor 1004 to implement various aspects or features for health management of magnetic storage media.
The media health manager 142 may be implemented independently or in combination with any suitable components or circuitry to implement aspects described herein. For example, the media health manager may be implemented as part of a DSP, processor/memory bridge, I/O bridge, graphics processing unit, memory controller, arithmetic Logic Unit (ALU), or the like. The media health manager 142 may also be provided integrally with other entities of the SoC 1000, such as with the processor 1004, memory 1006, storage medium interface, or firmware 1008 of the SoC 1000. Alternatively or additionally, the media health manager and/or other components of the SoC 1000 may be implemented as hardware, firmware, fixed logic circuitry, or any combination thereof.
As another example, consider FIG. 11 which illustrates an example storage medium controller 1100 in accordance with one or more aspects for health management of magnetic storage media. In general, storage medium controller 1100 enables computing device 102 to access the contents of a magnetic storage medium, such as an operating system, applications, or data for applications or other services. The storage medium controller may also write data of the computing device 102 to and read data of the computing device 102 from a magnetic storage medium associated with the controller.
In various aspects, the storage medium controller 1100, or any combination of components thereof, may be implemented as a storage drive controller (e.g., an HDD controller or HDD chipset), a storage medium controller, a NAS controller, a storage medium interface, a storage medium endpoint, a storage medium target, or a storage aggregation controller for magnetic storage media, solid state storage media, etc. (e.g., a hybrid SSD/HDD storage system). In some cases, storage medium controller 1100 is implemented similar to or with components of SoC 1500 (e.g., storage medium controller 136) as described with reference to fig. 10 or fig. 1. In other words, an instance of SoC 1000 may be configured as a storage medium controller (or subsystem of controllers), such as storage medium controller 1100 to manage magnetic storage media. In this example, the storage medium controller 1100 includes input-output (I/O) control logic 1102 and a processor 1104, such as a microprocessor, microcontroller, processor core, application processor, DSP, or the like. The storage media controller also includes a host interface 1106 (e.g., SATA, PCIe, NVMe, or Fabric interface) and a storage media interface 1108 (e.g., a magnetic media interface or head-disk assembly (HDA) interface) that enable access to host systems (or structures) and storage media, respectively. In this example, the storage media interfaces include separate instances of a spindle/VCM interface 1110, a preamplifier interface 1112, and a read/write channel interface 1114, enabling communication with the head-disk assembly and read channel of the media drive. As shown in FIG. 11, the storage medium controller 1100 may also include a servo control unit 138. In some cases, a servo control unit is operably coupled to the spindle interface 1110 and provides spindle or voice coil control for media drive operation.
In some aspects, the storage medium controller 1100 implements aspects of health management for magnetic storage media when managing or enabling access to storage media (e.g., media disks) coupled to the storage medium interface 1108. In this example, the storage medium controller 1100 includes a drive health manager 1016, which drive health manager 1016 may include medium health data 146 for multiple areas of the magnetic storage medium. In some aspects, the drive health manager 1016 receives read metrics or zone health scores from the read/write channel through the read/write channel interface 1114. Using read metrics or zone health scores from multiple media disks or media disk surfaces, the drive health manager 1016 can determine or update a health score (e.g., an overall health score of the HDD) for a magnetic storage medium (e.g., multiple disks) of the media drive. Thus, the drive health manager 1016 may monitor and track the respective health scores across multiple disks or surfaces of the magnetic storage medium of the media drive. Alternatively or additionally, the drive health manager 1016 may manage access to particular disks or surfaces based on their respective health scores. For example, if one disk of the media drive has a poor health score, the drive health manager 1016 may direct data access to other disks of the media drive (e.g., has a better health score) to improve the data reliability and availability (e.g., uptime) of the media drive. In some aspects, the firmware or logic of the processor 1104 and storage medium controller 1100 are implemented to provide various data writing or processing functions associated with health management for magnetic storage media.
The drive health manager 1016 of the storage medium controller 1100 may be implemented alone as shown, or in combination with the processor 1104, the read/write channel interface 1114, or the storage medium interface 1108. According to various aspects, the drive health manager 1016 may receive zone health scores from a read channel through the read/write channel interface 1114, signal metrics from a pre-amplifier through the pre-amplifier interface 1112, or electromechanical metrics from spindle and VCM components from the spindle/VCM interface 1110. The drive health manager 1016 may compile zone health scores, signal metrics, and/or electromechanical metrics, and calculate an overall media drive health score for the HDD in which the storage media controller 1100 is implemented. The media drive health score may indicate an overall amount of wear of internal components of the HDD or reliability of the HDD, enabling replacement of the HDD or migration (or redirection) of data to another HDD before the HDD fails. Examples of the components and/or entities, or corresponding functions, are described with reference to the various components or entities of the environment 100 of fig. 1 or the respective configurations shown in fig. 2-5. The drive health manager 1016 (or media health manager 142) may be implemented in whole or in part as processor-executable instructions maintained by the memory of the controller and executed by the processor 1104 to implement various aspects and/or features for health management of magnetic storage media.
Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific examples, features, or acts described herein, including the order in which they are performed.
Claims (15)
1. A method for managing health of a magnetic storage medium of a storage medium drive, the method comprising:
receiving a request to read a sector of the magnetic storage medium;
determining, with a read channel of the storage medium drive, a read metric for the sector of the magnetic storage medium that was read based on the request;
determining, from a plurality of regions of the magnetic storage medium, a region of the magnetic storage medium in which the sector resides, each region of the magnetic storage medium corresponding to a respective subset of tracks on a disk of the magnetic storage medium, each track comprising a plurality of sectors;
accessing a read metric for the region of the magnetic storage medium, the read metric for the region including respective read metrics for the sector being read and other sectors residing in the region of the magnetic storage medium;
updating the read metrics for the region of the magnetic storage medium based on the read metrics determined for the sector of the magnetic storage medium to provide updated read metrics for the region of the magnetic storage medium, the updating of the read metrics including updating the read metrics for the sector that is read based on the read metrics determined for the sector; and
A health score for the region of the magnetic storage medium in which the sector resides is determined based on the updated read metrics for the region of the magnetic storage medium using a neural network.
2. The method of claim 1, further comprising:
storing the health score for the region of the magnetic storage medium to a region information table; or (b)
Transmitting the health score for the region of the magnetic storage medium to a storage medium controller of the storage medium drive.
3. The method according to claim 1 or 2, wherein:
accessing the read metrics for the region of the magnetic storage medium includes: reading a corresponding average value of the read metrics from a region information table; and
updating the read metrics for the region of the magnetic storage medium includes: the respective average value of the read metrics is updated based on the read metrics determined for the sectors of the magnetic storage medium.
4. A method according to claim 3, wherein:
the respective average of the read metrics includes: a respective short-term average of the read metrics and a respective long-term average of the read metrics; and
Updating the read metrics for the region of the magnetic storage medium includes: the respective short-term average of the read metrics and the respective long-term average of the read metrics are updated based on the read metrics determined for the sectors of the magnetic storage medium.
5. The method according to claim 1 or 2, further comprising:
receiving an electromechanical quantity for the storage media drive from a spindle and voice coil motor assembly of the storage media drive; or alternatively
Receiving signal metrics for the storage medium drive from a pre-amplifier of the storage medium drive; and
determining a drive health score for the storage medium drive based on at least two of the health score, the electromechanical metric, or the signal metric for the region; or (b)
Wherein the health score for the region of the magnetic storage medium indicates a wear level, a degradation level, or a reliability level of the region of the magnetic storage medium; or (b)
Wherein the region of the magnetic storage medium corresponds to a subset of at least three subsets of tracks on a platter surface of a magnetic storage medium disk.
6. The method of claim 1 or 2, wherein the read metrics for the sectors of the magnetic storage medium include one or more of: unrecovered sector rate, average retry method rate, average off-track detection, average number of seek errors, average number of errors corrected by an error correction code ECC, average syndrome weights in ECC, average number of ECC iterations, average of Log Likelihood Ratio (LLR) Probability Density Function (PDF) statistics, average Mean Square Error (MSE) of front-end loop signals, average length of detected media defects, average length of analog-to-digital converter (ADC) saturation, average delta of adaptation parameters, or average of sync mark errors; or (b)
Wherein the neural network comprises a pre-trained neural network configured to: mapping the read metric of the read channel to one of at least three health score classifications to determine the health score for the region of the magnetic storage medium.
7. An apparatus, comprising:
an interface for communicating data with a host;
a magnetic disk of a magnetic storage medium for storing said data;
A read channel to read the data from a sector of the magnetic storage medium;
a neural network to determine a media health score; and
a media health manager configured to:
determining, with the read channel, a read metric for a sector of the magnetic storage medium from which the data was read;
determining, from a plurality of regions of the magnetic storage medium, a region of the magnetic storage medium in which the sector resides, each region of the magnetic storage medium corresponding to a respective subset of tracks on the disk of the magnetic storage medium, each track comprising a plurality of sectors;
accessing a read metric for the region of the magnetic storage medium;
updating the read metrics for the region of the magnetic storage medium based on the read metrics determined for the sector of the magnetic storage medium to provide updated read metrics for the region of the magnetic storage medium, the updating of the read metrics including updating the read metrics for the sector that is read based on the read metrics determined for the sector; and
a health score for the region of the magnetic storage medium is determined based on the updated read metrics for the region of the magnetic storage medium using the neural network.
8. The apparatus of claim 7, wherein the media health manager is further configured to:
storing the health score for the region of the magnetic storage medium to a region information table; or alternatively
Transmitting the health score for the region of the magnetic storage medium to a storage medium controller of the device; and/or
To access the read metrics for the region of the magnetic storage medium, the medium health manager is further configured to: reading a corresponding average value of the read metrics from a region information table; and
to update the read metrics for the region of the magnetic storage medium, the medium health manager is further configured to: the respective average of the read metrics is updated based on the read metrics determined for the sectors of the magnetic storage medium.
9. The apparatus according to claim 7 or 8, wherein the apparatus further comprises:
a spindle and voice coil motor assembly associated with the disk of the magnetic storage medium;
a preamplifier configured to: amplifying a read signal generated by a read head associated with the disk of a magnetic storage medium; and
The media health manager is further configured to:
receiving electromechanical metrics for the disk of the device from the spindle and voice coil motor assembly; or alternatively
Receiving signal metrics for the disk of the device from the preamplifier; and
a health score for the device is determined based on at least two of the health score, the electromechanical metric, or the signal metric for the region of the magnetic storage medium.
10. The apparatus of claim 7 or 8, wherein the read metrics for the sector of the magnetic storage medium comprise one or more of: unrecovered sector rate, average retry method rate, average off-track detection, average number of seek errors, average number of errors corrected by an error correction code ECC, average syndrome weights in ECC, average number of ECC iterations, average of Log Likelihood Ratio (LLR) Probability Density Function (PDF) statistics, average Mean Square Error (MSE) of front-end loop signals, average length of detected media defects, average length of analog-to-digital converter (ADC) saturation, average delta of adaptation parameters, or average of sync mark errors; and/or
Wherein the neural network comprises a pre-trained neural network configured to: mapping the read metric of the read channel to one of at least three health score classifications to determine the health score for the region of the magnetic storage medium.
11. A system on a chip (SoC) comprising:
a preamplifier interface to receive a signal corresponding to data read from a sector of the magnetic storage medium;
read channel logic to process the signal corresponding to the data read from the sector of the magnetic storage medium;
a hardware-based processor;
a memory storing processor-executable instructions that implement a media health manager in response to execution by the hardware-based processor for:
determining, with the read channel logic, a read metric for a sector of the magnetic storage medium from which the data was read;
determining, from among a plurality of regions of the magnetic storage medium, a region of the magnetic storage medium in which the sector resides, each region of the magnetic storage medium corresponding to a respective subset of tracks on a disk of the magnetic storage medium, each track comprising a plurality of sectors;
Accessing a read metric for the region of the magnetic storage medium;
updating the read metrics for the region of the magnetic storage medium based on the read metrics determined for the sector of the magnetic storage medium to provide updated read metrics for the region of the magnetic storage medium, the updating of the read metrics including updating the read metrics for the sector that is read based on the read metrics determined for the sector; and
a health score for the region of the magnetic storage medium is determined based on the updated read metrics of the region of the magnetic storage medium using a neural network.
12. The system-on-chip of claim 11, wherein the media health manager is further implemented to:
storing the health score for the region of the magnetic storage medium to a region information table; or (b)
Transmitting the health score of the region of the magnetic storage medium to a storage medium controller of the system-on-chip.
13. The system-on-chip of claim 11 or 12, wherein:
to access the read metrics for the region of the magnetic storage medium, the medium health manager is further implemented to: reading a corresponding average value of the read metrics from a region information table; and
To update the read metrics for the region of the magnetic storage medium, the medium health manager is further implemented to: the respective average value of the read metrics is updated based on the read metrics determined for the sectors of the magnetic storage medium.
14. The system-on-chip of claim 11, wherein the read metrics for the sectors of the magnetic storage medium comprise one or more of: unrecovered sector rate, average retry method rate, average off-track detection, average number of seek errors, average number of errors corrected by an error correction code ECC, average syndrome weights in ECC, average number of ECC iterations, average of Log Likelihood Ratio (LLR) Probability Density Function (PDF) statistics, average Mean Square Error (MSE) of front-end loop signals, average length of detected media defects, average length of analog-to-digital converter (ADC) saturation, average delta of adaptation parameters, or average of sync mark errors.
15. The system-on-chip of claim 11 or 12, wherein the neural network comprises a pre-trained neural network configured to: mapping the read metrics of the read channels to one of at least five health score classifications to determine the health score for the region of the magnetic storage medium.
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US201962799608P | 2019-01-31 | 2019-01-31 | |
US62/799,608 | 2019-01-31 | ||
US16/737,174 | 2020-01-08 | ||
US16/737,174 US11450348B2 (en) | 2019-01-31 | 2020-01-08 | Health management for magnetic storage media |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111508532A CN111508532A (en) | 2020-08-07 |
CN111508532B true CN111508532B (en) | 2023-11-03 |
Family
ID=69423067
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010076783.7A Active CN111508532B (en) | 2019-01-31 | 2020-01-23 | Health management for magnetic storage media |
Country Status (4)
Country | Link |
---|---|
US (1) | US11450348B2 (en) |
EP (1) | EP3690883A1 (en) |
KR (1) | KR20200095364A (en) |
CN (1) | CN111508532B (en) |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10734016B2 (en) | 2018-08-21 | 2020-08-04 | Marvell Asia Pte, Ltd. | Pulse-based writing for magnetic storage media |
US11061582B2 (en) | 2019-02-26 | 2021-07-13 | Marvell Asia Pte, Ltd. | Codeword interleaving for magnetic storage media |
US10971187B2 (en) | 2019-03-11 | 2021-04-06 | Marvell Asia Pte, Ltd. | Constant-density writing for magnetic storage media |
CN112506420A (en) * | 2019-09-16 | 2021-03-16 | 伊姆西Ip控股有限责任公司 | Method, apparatus and article of manufacture for managing scrubbing operations in a storage system |
CN112051969B (en) * | 2020-08-07 | 2021-10-08 | 东北大学 | Hierarchical storage data self-adaptive migration method based on deep reinforcement learning |
CN112214169B (en) * | 2020-09-28 | 2021-08-10 | 深圳大学 | Data storage device and stored data migration method |
WO2022081255A1 (en) * | 2020-10-13 | 2022-04-21 | Arris Enterprises Llc | Home network health metrics reporting |
US11899558B2 (en) * | 2021-01-15 | 2024-02-13 | Netflix, Inc. | Systems and methods for optimizing hard drive throughput |
CN112817523B (en) * | 2021-01-19 | 2021-09-07 | 置富科技(深圳)股份有限公司 | Storage medium reliability grade judging method and system, storage medium and equipment |
US12020733B2 (en) | 2022-04-18 | 2024-06-25 | Western Digital Technologies, Inc. | Sector metrics to estimate health of written data |
CN114822664B (en) * | 2022-05-12 | 2024-06-14 | 合肥兆芯电子有限公司 | Risk assessment method based on data priority, storage device and control circuit |
CN114860624B (en) * | 2022-05-23 | 2023-04-28 | 深圳市芯存科技有限公司 | Data erasing method and system based on Nandflash chip |
Citations (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0628766A (en) * | 1992-07-09 | 1994-02-04 | Fujitsu Ltd | Magnetic storage medium, magnetic writing/reading processor and its control method |
JPH10188451A (en) * | 1996-12-25 | 1998-07-21 | Toshiba Corp | Disk device, and failure predicting method based on defect in the device |
JP2001250292A (en) * | 2000-03-06 | 2001-09-14 | Fujitsu Ltd | Storage device and recording medium |
CN1347115A (en) * | 2000-09-11 | 2002-05-01 | 国际商业机器公司 | Rotating memory apparatus and controlling method therefor |
US6574754B1 (en) * | 2000-02-14 | 2003-06-03 | International Business Machines Corporation | Self-monitoring storage device using neural networks |
US6658201B1 (en) * | 1999-06-24 | 2003-12-02 | Sony Electronics, Inc. | Data storage device having multiple heads traveling in opposite directions for capacity and throughput optimization |
JP2006285519A (en) * | 2005-03-31 | 2006-10-19 | Hitachi Global Storage Technologies Netherlands Bv | Fault diagnostic method for data transfer system, data transfer system, and data storage device |
JP2006351134A (en) * | 2005-06-17 | 2006-12-28 | Fujitsu Ltd | Magnetic disk device and patrol diagnosing method |
CN102129397A (en) * | 2010-12-29 | 2011-07-20 | 深圳市永达电子股份有限公司 | Method and system for predicating self-adaptive disk array failure |
CN102800328A (en) * | 2011-05-25 | 2012-11-28 | 日立环球储存科技荷兰有限公司 | Artificial neural network application for magnetic core width prediction and modeling for magnetic disk drive manufacture |
US8953271B1 (en) * | 2013-05-13 | 2015-02-10 | Western Digital Technologies, Inc. | Disk drive compensating for repeatable run out selectively per zone |
US8953265B1 (en) * | 2014-07-24 | 2015-02-10 | Emc Corporation | Method and system for monitoring disk reliability with global disk scrubbing |
WO2015029163A1 (en) * | 2013-08-28 | 2015-03-05 | パイオニア株式会社 | Recording/reproducing apparatus, recording/reproducing method, and recording medium |
CN104932834A (en) * | 2014-03-21 | 2015-09-23 | 西部数据技术公司 | Data management of data storage device based on ultimate means areas |
GB201516823D0 (en) * | 2014-09-26 | 2015-11-04 | HGST Netherlands BV | Storage device health diagnosis |
US9189309B1 (en) * | 2013-09-25 | 2015-11-17 | Emc Corporation | System and method for predicting single-disk failures |
US9672853B1 (en) * | 2015-06-01 | 2017-06-06 | Marvell International Ltd. | Digital data driven position error signal for magnetic recording |
CN107861829A (en) * | 2017-11-06 | 2018-03-30 | 郑州云海信息技术有限公司 | A kind of method, system, device and the storage medium of disk failure detection |
CN107886992A (en) * | 2017-11-06 | 2018-04-06 | 郑州云海信息技术有限公司 | A kind of RAID method for detecting health status, system and relevant apparatus |
Family Cites Families (51)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4553178A (en) | 1983-01-18 | 1985-11-12 | Electronic Processors, Inc. | Magnetic recording system with peak shift compensation |
JPS60185258A (en) | 1984-03-02 | 1985-09-20 | Nec Corp | Information recording system of disc device |
US5420730A (en) | 1990-08-17 | 1995-05-30 | Moon; Ronald R. | Servo data recovery circuit for disk drive having digital embedded sector servo |
JPH0795056A (en) | 1993-05-10 | 1995-04-07 | Internatl Business Mach Corp <Ibm> | Variable frequency standard clock formation device |
JP2557203B2 (en) | 1993-12-27 | 1996-11-27 | インターナショナル・ビジネス・マシーンズ・コーポレイション | Fuzzy packing method and data storage system |
JPH07211004A (en) | 1994-01-20 | 1995-08-11 | Fujitsu Ltd | Magnetic disc apparatus and recording medium |
JP3456781B2 (en) | 1995-02-06 | 2003-10-14 | 富士通株式会社 | Demodulation circuit of magnetic recording / reproducing device |
JPH08287618A (en) | 1995-04-17 | 1996-11-01 | Matsushita Electric Ind Co Ltd | Recording and reproducing device |
KR100212982B1 (en) | 1996-05-22 | 1999-08-02 | 윤종용 | Constant density recording method and system |
JPH10255402A (en) | 1997-03-10 | 1998-09-25 | Fujitsu Ltd | Magnetic disk device |
US6023386A (en) | 1997-10-31 | 2000-02-08 | Cirrus Logic, Inc. | Fault tolerant sync mark detector for synchronizing a time varying sequence detector in a sampled amplitude read channel |
AU761402B2 (en) | 1998-08-04 | 2003-06-05 | Koninklijke Philips Electronics N.V. | Optical disc and apparatus for scanning the optical disc |
DE19852037A1 (en) | 1998-11-11 | 2000-05-18 | Thomson Brandt Gmbh | Magnetic tape device e.g. for use with computer or electronic camera, has driver circuit for recording head matrix employing common row and column return lines |
US6175319B1 (en) | 1999-05-20 | 2001-01-16 | Storage Technology Corporation | Power-tailored write-current method |
KR100611956B1 (en) | 1999-08-19 | 2006-08-11 | 삼성전자주식회사 | Error correction method and apparatus thereof |
US6894967B2 (en) | 2000-11-21 | 2005-05-17 | Yamaha Corporation | Optical disk recorder for writing data with variable density |
US6765736B2 (en) | 2001-06-29 | 2004-07-20 | Seagate Technology Llc | Write-safe condition verification apparatus and method for the write element in a disc drive |
US7088534B2 (en) | 2003-04-08 | 2006-08-08 | Marvell International Ltd. | Disk synchronous write |
US7236911B1 (en) | 2004-06-16 | 2007-06-26 | Western Digital Technologies, Inc. | Using a genetic algorithm to select a subset of quality metrics as input to a disk drive failure prediction algorithm |
JP2006031825A (en) | 2004-07-16 | 2006-02-02 | Sony Corp | Recording and reproducing control method, and recording and reproducing control apparatus |
KR100604892B1 (en) | 2004-08-10 | 2006-07-28 | 삼성전자주식회사 | Optical disc reproducing/rewriting system having defect detection circuit for managing defects of the re-writable disc and method thereof |
US7796355B2 (en) | 2005-12-01 | 2010-09-14 | Seagate Technology Llc | Reverse write current preshoot |
JP2007184063A (en) | 2006-01-10 | 2007-07-19 | Renesas Technology Corp | Nonvolatile semiconductor storage device |
JP2010218631A (en) | 2009-03-17 | 2010-09-30 | Toshiba Storage Device Corp | Error correction circuit and data storage device |
KR20110048776A (en) * | 2009-11-03 | 2011-05-12 | 삼성전자주식회사 | Method for processing defect of recording media and data storage apparatus and storage media applying the same |
US8687304B1 (en) | 2010-01-18 | 2014-04-01 | Marvell International, Ltd. | Sensor-aided patterned magnetic media synchronization systems |
US8508878B1 (en) | 2010-01-21 | 2013-08-13 | Marvell International Ltd. | Write clock phase tracking for magnetic recording device |
US8508879B1 (en) | 2010-01-21 | 2013-08-13 | Marvell International Ltd. | Write clock rephase for magnetic recording device |
CN104094349B (en) | 2011-12-05 | 2016-08-24 | 瓦里安半导体设备公司 | Write head, magnetic storage device and the method in magnetic storage device record data |
US9059737B2 (en) | 2013-01-11 | 2015-06-16 | HGST Netherlands B.V. | Disk drive with distributed codeword blocks |
US20140244926A1 (en) | 2013-02-26 | 2014-08-28 | Lsi Corporation | Dedicated Memory Structure for Sector Spreading Interleaving |
US8749911B1 (en) | 2013-03-11 | 2014-06-10 | Western Digital Technologies, Inc. | Disk drive accounting for fractional clock cycle when measuring reader/writer gap |
US8861117B2 (en) | 2013-03-15 | 2014-10-14 | Seagate Technology Llc | Data intersector gaps for data sector format |
US9094046B2 (en) | 2013-09-03 | 2015-07-28 | Lsi Corporation | Systems and methods for variable sector count spreading and de-spreading |
US8792197B1 (en) | 2013-09-20 | 2014-07-29 | Lsi Corporation | Storage device with driver controller providing pattern-dependent write functionality |
US8929013B1 (en) | 2013-11-08 | 2015-01-06 | Lsi Corporation | Storage system with pattern dependent write |
US9355677B2 (en) | 2014-01-30 | 2016-05-31 | Seagate Technology Llc | Preamplifier for adjustable write current |
US9281009B1 (en) | 2014-12-18 | 2016-03-08 | Western Digital Technologies, Inc. | Data storage device employing variable size interleave written track segments |
US9530447B2 (en) | 2015-02-27 | 2016-12-27 | Seagate Technology Llc | Adaptive HAMR power data storage device |
CN107179968B (en) * | 2016-03-09 | 2020-06-26 | 株式会社东芝 | Information storage device, failure prediction device, and failure prediction method |
US20170270963A1 (en) | 2016-03-17 | 2017-09-21 | Kabushiki Kaisha Toshiba | Magnetic disk device, data processing device and recording method |
US10268553B2 (en) | 2016-08-31 | 2019-04-23 | Seagate Technology Llc | Adaptive failure prediction modeling for detection of data storage device failures |
US10310749B2 (en) * | 2016-09-16 | 2019-06-04 | Netscout Systems Texas, Llc | System and method for predicting disk failure |
US10649655B2 (en) * | 2016-09-30 | 2020-05-12 | Western Digital Technologies, Inc. | Data storage system with multimedia assets |
US10276198B1 (en) | 2017-01-12 | 2019-04-30 | Marvell International Ltd. | Method and apparatus for multi-channel servo demodulation |
US10068609B1 (en) | 2017-02-23 | 2018-09-04 | Marvell International Ltd. | Variable frequency write pattern generation |
US10115415B1 (en) | 2018-03-23 | 2018-10-30 | Headway Technologies, Inc. | Magnetic pulse recording scheme in perpendicular magnetic recording |
US10734016B2 (en) | 2018-08-21 | 2020-08-04 | Marvell Asia Pte, Ltd. | Pulse-based writing for magnetic storage media |
US10832716B2 (en) | 2018-12-19 | 2020-11-10 | Marvell Asia Pte, Ltd. | Zone self servo writing with synchronized parallel clocks |
US11061582B2 (en) | 2019-02-26 | 2021-07-13 | Marvell Asia Pte, Ltd. | Codeword interleaving for magnetic storage media |
US10971187B2 (en) | 2019-03-11 | 2021-04-06 | Marvell Asia Pte, Ltd. | Constant-density writing for magnetic storage media |
-
2020
- 2020-01-08 US US16/737,174 patent/US11450348B2/en active Active
- 2020-01-13 KR KR1020200004183A patent/KR20200095364A/en not_active Application Discontinuation
- 2020-01-23 CN CN202010076783.7A patent/CN111508532B/en active Active
- 2020-01-30 EP EP20154549.8A patent/EP3690883A1/en active Pending
Patent Citations (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0628766A (en) * | 1992-07-09 | 1994-02-04 | Fujitsu Ltd | Magnetic storage medium, magnetic writing/reading processor and its control method |
JPH10188451A (en) * | 1996-12-25 | 1998-07-21 | Toshiba Corp | Disk device, and failure predicting method based on defect in the device |
US6658201B1 (en) * | 1999-06-24 | 2003-12-02 | Sony Electronics, Inc. | Data storage device having multiple heads traveling in opposite directions for capacity and throughput optimization |
US6574754B1 (en) * | 2000-02-14 | 2003-06-03 | International Business Machines Corporation | Self-monitoring storage device using neural networks |
JP2001250292A (en) * | 2000-03-06 | 2001-09-14 | Fujitsu Ltd | Storage device and recording medium |
CN1347115A (en) * | 2000-09-11 | 2002-05-01 | 国际商业机器公司 | Rotating memory apparatus and controlling method therefor |
JP2006285519A (en) * | 2005-03-31 | 2006-10-19 | Hitachi Global Storage Technologies Netherlands Bv | Fault diagnostic method for data transfer system, data transfer system, and data storage device |
JP2006351134A (en) * | 2005-06-17 | 2006-12-28 | Fujitsu Ltd | Magnetic disk device and patrol diagnosing method |
CN102129397A (en) * | 2010-12-29 | 2011-07-20 | 深圳市永达电子股份有限公司 | Method and system for predicating self-adaptive disk array failure |
CN102800328A (en) * | 2011-05-25 | 2012-11-28 | 日立环球储存科技荷兰有限公司 | Artificial neural network application for magnetic core width prediction and modeling for magnetic disk drive manufacture |
US8953271B1 (en) * | 2013-05-13 | 2015-02-10 | Western Digital Technologies, Inc. | Disk drive compensating for repeatable run out selectively per zone |
WO2015029163A1 (en) * | 2013-08-28 | 2015-03-05 | パイオニア株式会社 | Recording/reproducing apparatus, recording/reproducing method, and recording medium |
US9189309B1 (en) * | 2013-09-25 | 2015-11-17 | Emc Corporation | System and method for predicting single-disk failures |
CN104932834A (en) * | 2014-03-21 | 2015-09-23 | 西部数据技术公司 | Data management of data storage device based on ultimate means areas |
US8953265B1 (en) * | 2014-07-24 | 2015-02-10 | Emc Corporation | Method and system for monitoring disk reliability with global disk scrubbing |
GB201516823D0 (en) * | 2014-09-26 | 2015-11-04 | HGST Netherlands BV | Storage device health diagnosis |
US9672853B1 (en) * | 2015-06-01 | 2017-06-06 | Marvell International Ltd. | Digital data driven position error signal for magnetic recording |
CN107861829A (en) * | 2017-11-06 | 2018-03-30 | 郑州云海信息技术有限公司 | A kind of method, system, device and the storage medium of disk failure detection |
CN107886992A (en) * | 2017-11-06 | 2018-04-06 | 郑州云海信息技术有限公司 | A kind of RAID method for detecting health status, system and relevant apparatus |
Non-Patent Citations (1)
Title |
---|
分布式存储***中磁盘故障检测机制;刘榴;李小勇;;信息技术(05);全文 * |
Also Published As
Publication number | Publication date |
---|---|
TW202046297A (en) | 2020-12-16 |
KR20200095364A (en) | 2020-08-10 |
CN111508532A (en) | 2020-08-07 |
EP3690883A1 (en) | 2020-08-05 |
US20200251143A1 (en) | 2020-08-06 |
US11450348B2 (en) | 2022-09-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111508532B (en) | Health management for magnetic storage media | |
US11119660B2 (en) | Determining when to replace a storage device by training a machine learning module | |
US8645641B2 (en) | Intelligent storage device controller | |
CN113474838B (en) | Method, apparatus and system on chip for improving data reliability of magnetic storage medium | |
US9837115B1 (en) | Unequal error correction code in multi-track recording | |
US20080192378A1 (en) | Adaptive data-dependent noise prediction (ddnp) | |
JP2013255221A (en) | Data processing system with retained sector reprocessing | |
US20230020009A1 (en) | Health Management for Magnetic Storage Media | |
CN111344788B (en) | System level error correction coding allocation based on device population data integrity sharing | |
US10275309B2 (en) | Multi-layer integrated zone partition system error correction | |
US11237893B2 (en) | Use of error correction-based metric for identifying poorly performing data storage devices | |
US8743493B1 (en) | Fixed-point processing using quantization levels based on floating-point processing | |
TWI845597B (en) | Health management for magnetic storage media | |
US11593204B2 (en) | Fleet health management device classification framework | |
US10552252B2 (en) | Patterned bit in error measurement apparatus and method | |
US20180012627A1 (en) | Managing far and near track erasure by dynamic control of a write current parameter of a magnetic disk drive | |
JP2014182848A (en) | Information reproduction apparatus and information reproduction method | |
US8730606B1 (en) | Read channel error correction using multiple calibrators | |
CN115087962A (en) | Preemptive upgrade for full-span demotion | |
US10969969B2 (en) | Use of recovery behavior for prognosticating and in-situ repair of data storage devices | |
US11984910B1 (en) | Reinforcement learning-enabled low-density parity check decoder | |
US11341989B2 (en) | Read head stress reduction | |
US11386921B2 (en) | Read head stress reduction | |
Venkataraman | Circuit and system design for shingled magnetic recording |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |